The precise classification of histopathological images is crucial for diagnosing and treating cancer, yet the scarcity of labeled data often limits it. This study investigates the efficacy of using StyleGAN2-ADA data augmentation in histopathological image classification. In this work, we (i) evaluate the capability of StyleGAN2-ADA to generate realistic synthetic histopathological images, (ii) implement a data augmentation strategy using these images, and (iii) compare the performance of a binary classifier trained with and without the proposed augmentation. We trained a StyleGAN2-ADA model in a high-performance computing environment to generate high-quality synthetic images to augment histopathological datasets. We then trained binary classifiers using the augmented datasets for the PCam and IDC datasets and compared their performance with classifiers trained only with original data. Results showed a significant improvement in classifier accuracy, with a 5.9% increase in ROC (AUC) for the PCam dataset at 3% data availability and an 11.3% increase at 20% data availability. For the IDC dataset, the ROC (AUC) improved by 3.4% at 3% data availability and by 2.4% at 20% data availability. Notable enhancements were observed in cancer class metrics, particularly in low-data scenarios, demonstrating the effectiveness of StyleGAN2-ADA in improving classifier robustness and generalization. We conclude that StyleGAN2-ADA is an effective tool for generating high-quality synthetic histopathological images and that the proposed data augmentation strategy substantially enhances classifier performance. Thus, we improved the robustness and generalization of classification models in this critical medical field. Furthermore, it highlights the importance of HPC in accelerating deep learning research applied to complex medical problems.